一种快速稳定的倾斜文档校正方法

摘要

在文档扫描数字化过程中会出现倾斜,依据文档图像自身的线条和文字行特征,提出了一种快速而稳健的文档校正算法.该算法首先对文本进行数学形态学处理、边缘检测,然后利用直线拟合技术得到直线,从中筛选出有代表性的直线,通过直线与主轴方向的夹角检测出角度,最后对图像进行旋转校正.通过实验验证以及与目前具有相关代表性的方法对比,论证了本算法对带有表格、图像、手写体,以及各种语言的文本都具有很好的校正效果,具有良好的有效性和可靠性.%In the process of document scanning and digitization, a skew angle of scanned document can occur easily. A highly effective method is proposed based on the lines and lines of text features on the document images. Firstly, mathematical-morphological processing and edge detection are used to various complex texts. Some straight lines are gotten by linear fitting and some representative straight lines are selected from the obtained straight lines. The skew angle is determined according to the one between the axis direction and the straight lines detected. At last, the document image is rectified by the detected skew angle. Experimental results show that the proposed algorithm has good effectiveness and reliability with forms, images, script, and various language texts by comparison with the state-the-art methods.